Constant power-mode optical amplifiers are becoming increasingly important as high bit-rate (>10 Gbps), reconfigurable optical transmission systems with complex networking architectures, including reconfigurable optical add/drop multiplexers (ROADM), become more popular. The addition of such ROADM components adds loss and increases the need for terminal-EDFAs immediately preceding receivers beyond the more traditional in-line EDFA's as shown in FIG. 1. High bit-rates (40 Gbps and greater) have more stringent optical signal to noise ratio (OSNR) requirements than low bit-rates and also require the deployment of per-channel tunable dispersion compensators as shown in FIG. 1, further increasing the need for terminal-EDFA's. Due to dynamic channel add/drop events from the aforementioned increasing use of ROADMs, the input power levels to terminal-EDFAs can vary in a big range. Transient overshoots at the terminal-EDFA output resulting from a sudden input power surge can cause irreparable damage to the receiver. This receiver input power limit serves to lower the maximum acceptable output power of terminal EDFA's while the Optical Signal to Noise Ratio (OSNR) constraint from high bit-rate serves to raise the minimum acceptable output power of the terminal EDFA. Hence, fast overshoot suppression is a useful feature in constant power-mode EDFAs. The OSNR requirements typically prevent users from simply operating the terminal-EDFAs at output power levels low enough to permit the application of conventional constant gain-mode EDFA control techniques.
Although constant-gain controlled EDFAs, wherein the inherent gain drift of the amplifier is compensated by adjustments in the pump power, are widely deployed, very fast constant-power controlled EDFAs present a different challenge, and solutions are still being sought. Simple and low-cost solutions are always desirable but challenging. A key problem is to suppress the output power surge that results from an input power transient in a short enough time to ensure that the integrated energy hitting a photodetector is substantially less than that required to cause overload or damage. If any EDFA is added upstream of a PIN photodetector it is important to suppress power transients to avoid overload or damage to the photodetector.
One approach to addressing this problem is described in U.S. Pat. No. 6,757,099. The solution described in this patent involves the use of a feedfoward and feedback control algorithm to do the power transient suppression. However, this method does not provide sufficiently fast response to very fast transient rise times.
Another approach is described in U.S. Pat. No. 5,187,610, which describes the use of a lumped loss element at the output of an EDFA to increase the noise figure of merit. However, it is not evident that this solution adequately addresses issues of fast transient performance.
Known prior art transient control schemes operate in some fashion by raising and lowering the pump power in response to a change of the input power. These are limited by the characteristic response time of the gain medium.
Thus there is a need for improvements in the suppression of fast input power transients.